[11] Patent Number: 4,613,943 [45] Date of Patent Sep. 23, 1986

United States Patent [191 Miyake et al.

S m m MU 6% s mN .mE RM P s U Q B [54] OPERATION TEACHING METHOD AND

APPARATUS FOR INDUSTRIAL ROBOT

_ Re. 30,016 5/1979 Hohn [75] Inventors: Norihisa Miyake, lbaraki; Akira 3,909,600 9/1975 Hohn _ _

Tsuchihashi, Nagareyama; Osamu 3,920,972 11/1975 Corwin, Jr. et a1. .. Fujiwara, Funabashi; Yasuhiro 4,086,522 4/1978 Engelberger et a1. .

Hashimoto, Ibamki; Yutaka 4,495,588 1/1985 Nio et al. Maruyama, Ashikaga, all of Japan 4,517,653 5/1985 Tsuchihashi et a1. ..

Primary Examiner—Joseph Ruggiero Attorney, Agent, or Firm-Antonelli, Terry & Wands

[57] [73] Assignee: Hitachi, Ltd., Tokyo, Japan

ABSTRACT

[21] Appl. No.: 599,472 An operation teaching method and apparatus for an industrial robot adapted to be successively moved to

. and set at different positions along an objective struc [22] Filed‘ Apr‘ 12’ 1984 ture to conduct a predetermined operation on working

objects of the same con?guration on the objective struc [30] Foreign Application Priority Data ture to which objects the different positions correspond,

respectively. The data taught at an initial position is Apr. 13, 1983 [JP] Japan 58-63593 corrected throughacoordinate transformation between

a coordinate system ?xed on a working object and a coordinate system assumed on the robot, and the cor [51] Int. 01.4 G06F 15/46; GOSB 121/42

364/513 rected data are reproduced and used as the operation [52] us. c1. .. ; 318/568; 901/3; 901/42

364/513, 191-193, 364/191 data for the second and the following working posi

tions. 3

[58] Field of Search 364/474, 478; 318/568; 901/1-7, 41, 42, 50;

414/730 18 Claims, 7 Drawing Figures

US. Patent Sep.23, 1986 Sheetl 0f5 4,613,943

- - IIIIIIIIIAWIIIIJQ < s N A E

4,613,943 Sheet 2 of 5 US. Patent Sep.23,1986

_US. Patent Sep.23, 1986 Sheet4of5 4,613,943

FIG. 5

FIG.6

US. Patent Sep.23, 1986 SheetSofS. 4,613,943

4,613,943 1

OPERATION TEACHING METHOD AND APPARATUS FOR INDUSTRIAL ROBOT

BACKGROUND OF THE INVENTION

The present invention vrelates to a method of and apparatus for teaching an operation to an industrial robot and, more particularly, to an operation teaching method and apparatus suitable for industrial robots which are intended to be moved to a plurality of posi tions and to perform the same task at these positions.

Generally, industrial robots are ?xed and perform tasks on works which are disposed within the reaches of the industrial robots. In the case where the works are of the same con?guration, it is preferred from the view point of the working ef?ciency of the robot and the labour saving effect that the content of the operation taught to the robot for one of the works is memorized and reproduced at each time of operation to enable the robot to perform the same task on a plurality of works of the same con?guration. This control method is gen erally referred to as a “teaching and playback control method” and a typical example of this method is dis closed in U.S. Pat. No. 3,920,972.

In recent years, the kinds of industrial robots are so diversi?ed and there is an increasing demand for indus trial robots which can perform tasks on the works in stalled on large structures which are dif?cult to move or on the works installed inside of box-shaped struc tures. In order to cope with this demand, it is necessary to take a suitable measure for moving the robot to the position of the work and to ?x the same at this position. In the system in which the industrial robot has to be moved to different positions, it is impossible to adopt the above-mentioned teaching and playback control method which is designed for an industrial robot which is intended to perform a task at ?xed position. Namely, assuming that the industrial robot ?rst performs the task on a work at a ?rst position and then moves to a second position where it performs the same task on the new work, it is quite dif?cult to realize in the second position the same positional relationship between the robot and the work as that attained in the ?rst position. It is, there fore, impossible to effect the desired task on the work at the second position by reproducing the content of the operation taught in the ?rst position. Consequently, it is necessary to teach the content of the operation again to the robot after the robot is moved to the second posi tion.

SUMMARY OF THE INVENTION

Accordingly, an object of the invention is to provide an operation teaching method and apparatus for an industrial robot which is improved to remarkably shorten the length of time required for teaching the content of the operation when the robot has been moved to a new position, even when the positional relationship between the robot and the work in the new position differs from that in the previous position of the robot. Another object of the invention is to provide an oper

ation teaching method and apparatus which can widen the scope or application of the industrial robot to works having the same con?guration. To these ends, according to one aspect of the inven

tion, there is provided an operation teaching method for an industrial robot which is adapted to be successively moved to and set at different positions along an objec

20

25

35

45

55

60

65

2 tive structure to conduct a predetermined operation on working objects of the same con?guration on the objec tive structure to which objects said different positions correspond, respectively, the method comprising: teaching, when the industrial robot is set at a ?rst posi tion corresponding to a ?rst working object, positions of base references and a locus of the robot operation assumed on the ?rst working object, and storing such information as coordinate values given in terms of a ?rst robot coordinate system assumed on the robot set in the ?rst position; teaching, when the industrial robot is set at a second position corresponding to a second working object, positions of comparison references assumed on the second working object at positions corresponding to the base reference positions, and storing such infor mation as coordinate values given in terms of a second robot coordinate system assumed on the industrial robot set at the second position; calculating correlation infor mation describing the relationship between the ?rst robot coordinate system and the second robot coordi nate system from the coordinate values of the base ref erences in terms of the ?rst robot coordinate system and the coordinate values of the comparison references in terms of the second robot coordinate system; and cor recting, by using the aforementioned correlation infor mation describing the relationship between the ?rst robot coordinate system and the second robot coordi nate system, the coordinate values representing the locus of the robot operation in terms of the ?rst robot coordinate system to determine coordinate values rep resenting the locus of the robot operation in terms of the second robot coordinate system. According to another aspect of the invention, there is

provided an operation teaching apparatus for an indus trial robot adapted to be successively moved to and set at different positions along an objective structure to conduct a predetermined operation on working objects of the same con?guration on the objective structure to which objects the different positions correspond, re spectively, the apparatus comprising: memory means for storing the content of a robot operation on the working objects taught to the industrial robot; ?rst computing means for calculating correlation between a ?rst robot coordinate system assumed on the industrial robot set at a ?rst position corresponding to a ?rst working object and a second robot coordinate system assumed on the industrial robot set at a second position corresponding to a second working object from coordi nate values of base references assumed on the ?rst working object and coordinate values of comparison references assumed on the second working object, said coordinate values of the base references being stored in said memory means in terms of the ?rst robot coordi nate system, and said coordinate values of the compari son references being stored in said memory means in terms of the second robot coordinate system; and sec ond computing means for conducting, using the fore mentioned correlation information between the ?rst and second robot coordinate systems, a correcting computa tion of coordinate values representing the taught opera tion of the robot and stored in said memory means in terms of the ?rst robot coordinate system to determine coordinate values representing the taught operation of the robot in terms of the second robot coordinate sys tern. These and other objects, features and advantages of

the invention will become clear from the following

4,613,943 3

description of the preferred embodiments taken in con junction with the a'ccompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a perspective view showing, partly in sec tion, an object which is to be processed by an industrial robot to which the present invention is applied; FIG. 2 is a perspective view of an industrial robot to

which the present invention is applied; FIG. 3 is an illustration of the principle of operation

teaching method in accordance with the invention; FIG. 4 is a block diagram of an embodiment of opera

tion teaching apparatus of the invention; and FIGS. 5 to 7 are illustrations of the principle of differ

‘ ent embodiments of operation teaching method in ac cordance with the invention.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

Preferred embodiments of the invention will be de scribed hereinunder with reference to the accompany ing drawings.

Referring ?rst to FIG. 1, an industrial robot R is intended for performing tasks on an objective structure 1 which is a box-type structure. More specifically, the objective structure 1 has a plurality of portions to be worked (referred to as “working objects”, hereinunder WM to W1C, to W2C and W3A to W3C. In order to per form the tasks on these working objects, the robot R is adapted to be moved successively to positions P, Q and V in the vicinity of working objects W1 3 and W1C. An example of the robot R is shown in FIG. 2. Namely, the robot R may be of the type having a turret 2, an upper arm 3 swingably mounted on the turret 2, a fore arm 4 swingably attached to the end of the upper arm 3, a wrist 5 swingably attached to the end of the fore arm 4, and a welding torch 6 provided on the wrist 5.

In operation, the robot R performs the tasks on the objective structure 1 in the following manner. Namely

‘ when the robot R takes the position P in the vicinity of the working object W1 A, the content of the operation to be performed on the working object WM is taught to the robot R, so that the robot R performs the taught operation on the working object W1 ,4. Then, the robot R is moved to the position Q in the vicinity of the work ing object W13 corresponding to the working object WM, so as to perform the same operation on the work ing object W11; as that performed previously on the working object W1A. In this manner, the robot R per forms the expected operations successively on the working objects W1A to W1C, WZA to W2C and W3,4 to W3C on the objective structure 1. According to the invention, when the robot R is

moved to positions different from the ?rst position where the content of the operation is taught to the robot, the content of the operation taught in the ?rst position is amended in accordance with the new posi tional relationships between the robot R and the work ing objects W1 1;, W1cso that the robot can perform the expected task‘ through teaching of only a few points, without necessitating reteaching of the entire operation in the new positions.

In order to make understood the teaching method in accordance with the invention, the principle of this teaching method will be described hereinunder with speci?c reference to FIG. 3. A ?xed coordinate system 0A, XA, YA and 2,4 is assumed as shown in FIG. 3 on the working object W1A which is to be processed by the

20

25

30

35

40

45

50

60

65

4 robot R installed or set at the position P shown in FIG. 1, while a corresponding coordinate system op, xp, yp and zp is assumed on the robot R. Similarly, a ?xed coordinate system 01;, X 3, Y3 and Z3 is assumed on the working object W13 which is to be processed by the robot R set at the position Q, and a corresponding coor dinate system 0Q, XQ, yQ, ZQ is assumed on the robot R. These coordinate systems assumed on the robot R cor respond to the coordinate system 0, x, y, z on the robot R shown in FIG. 2. The teaching of the content of the operation is made

on the robot R in the position P with respect to the working object W14. The point data obtained through the teaching, therefore, is based on the coordinate sys tem op, xp, yp, zp on the robot R located at the position

In order to move the robot R to the next position Q to enable the robot to perform the same operation on the working object W11; corresponding to the working position WM, it is necessary to express the point data, which has been obtained in the position P, in terms of the coordinate system 0Q, xQ, yQ, zQ assumed on the robot R in the position Q. As stated before, the content of the operation to be

performed by the robot on the working object W13 is identical to that performed by the robot on the working object W14. Therefore, the coordinate values (X,,, Y,,, Z”) of any desired point A on the working object WM in terms of the coordinate system 0,4, XA, YA, ZA coin cide with the coordinate values (Xb, Yb, 21,) of the corresponding point B on the working object W13 in terms of the coordinate system 03, X3, Y3 and Z3.

It is assumed here also that the coordinate values of the point A in terms of the coordinate system op, xp, yp, zp assumed on the robot R in the position P are ex pressed as (x_,,, yp, Zp), while the coordinate values of the point B in terms of the coordinate system 09, xQ, yQ, 2;; are given as (xq, yq, zq). It will be understood that, if a suitable equation is given for the transformation be tween the coordinate values (xp, yp, 21,) of the point A and those (xq, yq, zq) of the point B, it will become possi ble to enable the robot to perform the expected opera tion regardless of the set position of the robot R with respect to the working object W13, by amending the taught data by means of the transforming equation and then reproducing the amended content of the teaching. An explanation will be made hereinunder as to how

the transformation of the coordinate values mentioned above is conducted in a manner explained hereinunder. Namely, the coordinate values of any desired point A shown in FIG. 3 are expressed as (X,,, Y,,, Za) in terms of the coordinate system 0,1, XA, YA, Z; and as (xp, y_,,, Zp) in terms of the coordinate system op, Xp, yp, zp. The following relationship exists between these two coordi nate systems.

(1)

l 1

Where, T,,.] represents the transformation matrix for the transformation between two coordinate systems.

Similarly, the coordinate values of the point B corre sponding to the point A are expressed as (Xb, Yb, 2],) in terms of the coordinate system 03, X3, Y3, Z3 and as (xq, yq, Zq) in terms of the coordinate system 09, x9, yQ,

4,613,943 5

2g. The following relationship exists between these two coordinate systems.

(2)

l l

where, Twz represents the transformation matrix for the transformation between two coordinate systems.

Since the point A in the coordinate system 04, X4, Y4, 2,; corresponds to the point B in the coordinate system 03, X B, Y B, Z5, the following relationship exists between the coordinate values of these two points A and B.

(3) x1, 1000 X, Y1,_0l00 Y, z,_0o10 z, 1 00011

Therefore, the following equation (4) is derived from the equations (1), (2) and (3), as an equation which de termines the relationship between the coordinate values of the point A in the coordinate system op, xp, Yp, 2p and the coordinate values of the point B in the coordi nate system 0Q, xQ, yQ, 2g.

(4)

l l

where, Twg-l represents the inverse transformation matrix which is inverse to the transformation matrix Twg. The transformation matrices TW1 and Tm are ex

pressed by the following equations (5) and (6).

(5) 111 "111 "11 — X10

T _ 112 "112 "12 — y1o wl — I

13 m1; '113 —- Z10

0 O O l

(6) I21 "121 "21 — X20

Twz : $22 "122 "22 — .V20 23 "123 "23 — Z20

0001

Where (111, H111, I111), (112, H112, 1112) and (113, H113, I113) represents the direction cosines of the axes of the coor dinate systems op, xp, yp, zp on the axes of the coordi nate System 0A, XA, YA, ZA and (121, 11121, I121), (122, 11122, mg) and (123, m23, n23) represent the direction cosines of the axes of the coordinate system 09, rag, yQ, 29 on the axes of the coordinate system 01;, X3, YB, 23. A rela tionship expressed by the following equation (6A) is also established.

(6A)

121 122 123 121 X20 + 122 no + 123 Z20 Tw2_1 : "121 "122 "I23 "I21 X20 + M22 .1120 + "123 Z20

"21 v "22 I123 n21 x20 + 1122 no + "23 Z20 O O 0 1

30

35

45

50

55

60

65

6 In the equations (5) and (6) above, the coordinate

values (x10, ylo, Z10) represent the position of the origin of the coordinate system 0,4, X4, Y4, Z; on the coordi nate system 0 p, xp, yp, zp, while (x20, yzo, 220) represents the position of the origin of the coordinate system 03, X3, Y3, Z5 on the coordinate system 0Q, xQ, yQ, ZQ. The following relationships (7) are derived from the

nature of the direction cosine, in connection with the equations (5) and (6).

1112 + "1112 + "112 = 1,1112 + 1122 + 1,32 = l (7)

1122 + "1122 + m2 = 1. mm2 + "1122 + m32 = 1

1132 + "1132 + "132 = 1, "112 + "122 + m2 =

where, i is an integer which is l or 2. As will be understood from the foregoing descrip

tion, it is possible to effect the transformation of the coordinate values of the point on the coordinate system 0A, XA, YA, 2,, from the expression in terms of the coordinate system op, xp, yp, zp into an expression in terms of the coordinate system 0Q, xQ, yQ, zQ, provided that the transformation materials Twl and Twz are given.

In order to make the possibility of the above-men tioned transformation understood more clearly, an ex planation will be given hereinunder employing another way of stating the expression. Any desired point A* is assumed on the coordinate

system 0,4, X4, Y4, ZA. Then, a coordinate system 04*, X,4*, Y,4*, Z,4* is assumed to have the origin coinciding with the point A“. The transformation matrix between these two coordinate systems is expressed by TH. Simi larly, a coordinate system 03*, X3"‘, Y3“, Z5“ is as sumed on the coordinate system 03, X3, Y3, Z3. The transformation matrix for transformation between these coordinate systems is also expressed by TH. Represent ing the transformation matrix for transformation from the coordinate system 0;, xp, yp, Zp to the coordinate system 0/1"‘, X,,*, Y,;*, Z4‘ by Tm, the transformation matrix Twl is given by the following equation (8).

Similarly, representing the transformation matrix for the transformation from the coordinate system 0Q, XQ, yQ, zQ to the coordinate system 03*, X3“, Y3", Z3‘ by Tm, the transformation matrix Twz is expressed as fol lows.

Therefore, Twz" 1-Tw1 appearing in the equation (4) is rewritten as follows.

711721 ' Twl 131- T132‘ - TR1- TH <10)

T221- Tm

4,613,943 7

The equation (10) above suggests that any coordinate system can be assumed on the working object. Namely, it is possible to transform the expression in terms of the coordinate system op, xp, yp, zp assumed on the robot R in the position P to the expression in terms of the coor dinate system 0Q, xQ, yQ, zQ assumed on the robot R in the position Q, by a process which includes the steps of assuming any desired coordinate system 0,4“, XA“, YA“, Z,4* on the working object W1 ,4, determining the trans formation matrix TRl for transformation between the coordinate system 0,4“, X,;*, YA", Z,4* and the afore mentioned coordinate system Op, xp, yp, zp, assuming a coordinate system 03*, X5“, YB“, 23* corresponding to the coordinate system 0,;, X4, Y4, 2,; on another working object W1 3, and determining the transforma tion matrix TR; for the transformation between this coordinate system 05*, X3", Y3“, Z3“ and the afore mentioned coordinate system oQ, xQ, yQ, 29. An embodiment of the operation teaching apparatus

of the invention, making use of the principle explained hereinbefore, will be described hereinunder with spe ci?c reference to FIG. 4. A reference numeral 101 designates an operation path

memory device which is adapted to store base working data concerning the path or locus of operation of the robot R set at a base set position such as the position P

I shown in FIG. 3. This basic working data is given in the form of numerical data which in turn is determined by a computer or the like means from the operation taught

~ to the robot R or the design data. A reference numeral 102 designates a base reference point data memory de vice adapted to store base reference point data which determines the positional relationship between the working object and the robot R in the prescribed base set position. The base reference point data is inputted through, for example, teaching of the operation. The number of base reference point data can be increased or decreased in accordance with the content of correction ‘required to compensate for the positioning error of the -.robot which will be described later. The base reference point data is obtained, for example, in relation to the point A shown in FIG. 3. A comparison reference point data memory device 103 is adapted to store comparison reference point data which determines the positional relationship between the working object and the robot R when the robot R is set at each of the other positions (referred to as “correction set position”). This compari son reference point data is obtained in relation to, for example, the point B shown in FIG. 3. A reference numeral 104 designates a device for computing the transformation matrix Twl at the base set position in accordance with the equation (1) from the base refer ence point data delivered by the device 102. A refer ence numeral 105 designates a device for computing the transformation matrix Twz at the correction set position in accordance with the equation (2) from the compari son reference point data delivered by the device. A reference numeral 106 designates a device for comput ing the transformation matrix Twg- l-Twl for the correc tion of set position error in accordance with the equa tion (4) mentioned before from the results of computa tions performed by the devices 104, 105. A reference numeral 107 designates a device for correcting the oper ation path of the robot R delivered by the device 101, by making use of the set position error correction trans formation matrix Twrl-Twi delivered by the device 106. The corrected operation path data provided by this computing device 107 represents the operation path

20

25

30

40

45

55

65

8 data required for the robot R in the correction set posi tion. This data is stored in a memory device 108 and is successively forwarded to the operation instruction generating circuit 109 of the robot. The operation in struction generating circuit 109 drives the robot R in accordance with the aforementioned data to effect the operation on the working object at the correction set position. The computing devices 104 to 106 perform respective

computing operations only once at each time the robot R is moved from one to the next position. On the other hand, the computing device 107 effects the correction computation for all points on the operation path. There fore, if the operation speed of the computing device 1-7 is sufficiently high as compared with the operation speed of the robot R, it is possible to omit the memory device 108 because the computation performed by the computing device 107 can be made in real time during the operation of the robot R. An embodiment of the operation teaching method in

accordance with the invention will be explained herein under with reference to FIG. 5. In FIG. 5, the same reference numerals are used to denote the same parts or members as those used in FIG. 3. Referring to FIG. 5, reference points A1, A2 and A3 are assumed on any desired positions on the working objects WM of the objective structure 1 for the robot R set at the position P. These reference points A], A2, A3 will be referred to as “base reference points”, hereinunder. On the other hand, reference points B1, B2, B3 are assumed at any desired position on the working object W1 3 of the ob

' jective structure 1 for the robot R set at the position Q. These reference points B1, B2, B3 will be referred to as “comparison reference points”, hereinunder. The coor dinate values of the comparison reference points B1, B2, B3_on the coordinate system 03, X5, Y3 and Z3 coincide with the coordinate values of the base reference points A], A2, A3 on the coordinate system 0,4, XA, YA and ZA.

In operation, at ?rst the robot R is set at the position P adjacent to the working object W1 A. The positions of the three base reference points A1, A2, A3 on the work ing object WM of the objective structure 1, as well as the path of the working operation along the working object WlA are taught by the robot R. The data con cerning the base reference points A1, A2, A3 are stored in the base reference point data memory device 102 shown in FIG. 4. On the other hand, the taught data concerning the operation path along the working object W1 A is stored in the operation path memory device 101. Although in this embodiment the base reference points A1, A2, A3 are assumed at any desired positions on the working object W14, this is not exclusive and they may be set on the taught working operation path. In this case, needless to say, the comparison reference points B1, B2, B3 also are assumed on the working operation path. The computing device 104 performs the following computation in accordance with the base reference point data derived from the memory device 102. Namely, the coordinate values of the base reference point A1, A2, A3 on the coordinate system op, xp, yp, zp assumed on the robot R are expressed as follows.

(11)

4,613,943 On the other hand, the coordinate values of the base

reference points A], A2, A3 on the coordinate system 0,4“, X4“, Y4“, Z,4* assumed on the working object W1 ,4 are expressed as follows.

The coordinate values shown in the equations (11) and (12) above are expressed as follows, using the equa tion (1) and (5) mentioned before.

(Xm, YAlv ZAl) (XAL YAL 2A2) (XAz, YAs- 2A3)

(12)

There are 21 (twenty-one) unknowns in the equation (13) ?bOVeI namely, XAi, YAl, Z141, X43. YAz, 2A2. XA3, YAs, ZA3, l1, 12, 13, m1, m2, m3, 111, n2, n3, X10, yloand Z10 Among these 21 unknowns, the following 15 unknowns are independent in view of X4], YAl, 2A1, XAZ, YAZ, ZA2, XA3, YAs, 2A3, 11, 12, m1, X10. Y10 and Z10 The number of the unknowns can be reduced to 9

(nine) by adopting the following conditions in the coor dinate values of the base reference points A1, A2, A3 on the coordinate system 0,4, XA, YA, 2,4 as given by the equation (2).

Thus, the equation (13) above includes 9 (nine) equa tions. Since these equations are independent of each other, they are soluable by solving simultaneous equa

' tions. That is, the transformation matrix Twl is obtain able.

Then, the robot R is moved to and set at the position Q shown in FIG. 5, and the aforementioned comparison reference points B1, B2, B3 are taught to the robot. The data thus taught is stored in the memory device 103 shown in FIG. 3. The computing device 105 then solves the simultaneous equations using the data concerning the comparison reference points B1, B2, B3 in the same manner as described before, thereby to obtain the trans formation matrix Twz. Then, using this transformation matrix Twz and the transformation matrix Twl from the computing device 104, the computing device 106 shown in FIG. 3 computes the value of the transformation matrix Tw2-1-Tw1. Then, using the transformation ma trix Twr l-Tw1. Then, using the path for the robot R in the position P derived from the memory device 101 shown in FIG. 3, the computing device 107 performs a correction computation to determine the corrected operation path for the robot in the position Q. In accor dance with the information concerning the corrected operation path, the robot R performs the operation on the desired portion of the working object correspond ing to the robot position Q. From the foregoing description, it will be understood

that, in the operation at the second and other succeed ing positions, it is not necessary to teach the working operation path to the robot. Namely, the robot can

15

25

30

35

40

45

55

60

65

10 perform the operation on the working object of the same shape as that in the position P only by being taught the comparison reference points B1, B2 and B3. A second embodiment of the operation teaching

method of the invention will be described hereinunder with reference to FIG. 6.

In most cases, the robot to which the invention is applied is situated on a horizontal plane. Even if this requirement is not met, the robot generally takes con stant positions relative to a plurality of working objects of the same con?guration. In the case where the robot is intended for welding in particular, the quality of the work is largely affected by the welding posture. It is, therefore, possible to impose predetermined conditions for the postures of the working object and the working posture of the robot. In the case of the robot R shown in FIG. 2, the shaft 20 of the turret 2 is kept in a constant posture even when the robot R is moved to and set at different positions. Referring to FIG. 6, base reference points A] and A2 are assumed on the working object W1 ,4 for the robot R set at the position P. These base reference points A1 and A; may be selected such that the straight line interconnecting these points is not par allel to the turret shaft 20 of the robot R. On the other hand, the comparison reference points B1 and B2 are assumed on the working object W13 for the robot R set at the position Q, so as to correspond to the points A] and A2. When the robot is moved from the position P to the position Q, the working operation path of the robot taught at the position P is corrected using the base refer ence points A1 and A2 and the comparison reference points B1 and B2, so that the robot R in the position Q can perform on the working object the same operation as that performed by the robot R in the position P, as will be fully understood from the following description.

Relationships similar to those expressed by the equa tions (11) and (12) apply to the base reference points A1 and A2. A coordinate system 0,1, XA, YA, 2,4 is selected such that the origin of this coordinate system coincides with the point A1 and that the axis zp, i.e. the turret shaft 20, is parallel to the ZA axis, with the point A2 included by the plane de?ned by the axes X4 and 24.

Consequently, the following conditions are estab lished.

At the same time, the following conditions are met.

113=m13=n11=n12=0

In view of the conditions such as I112+m112= 1, l112+l122=l and n112+m122=1; there are 6 (six) un knowns 111, x10, ylo, Z10, XAZ and 2,42. On condition of i=1 or 2 in equation (13), the number of the equations is 6 (six) so that these unknowns can be determined. It is, therefore, possible to obtain the aforementioned trans formation matrix Twl. By moving the robot R to the position Q and teaching

it the comparison reference points B1, B2 corresponding to the base reference points A], A2, it is possible to obtain the transformation matrix Twz between the coor dinate system assumed on the robot R and the coordi nate system de?ning the points B1, B2. Using this trans

4,613,943 11

formation matrix Tm together with the aforementioned transformation matrix TW1, it is possible to obtain the transformation matrix Tm-l-Tm for correcting the operation path of the robot R in the position Q. Accord ing to this second embodiment, it is possible to reduce the number of the reference points as compared with the ?rst embodiment. A third embodiment of the operation teaching

method of the invention will be described hereinunder. This third embodiment is a more simpli?ed form of

the second embodiment. Namely, if two out of three of the axes of the coordinate system on the robot R are parallel to two out of the three axes of the coordinate system assumed on the working object, it is possible to make the transformation by a translational movement of one of the coordinate systems. This can be achieved by using only one base reference point on the working object for the robot set at one position and a corre sponding comparison reference point on the other working object at the position corresponding to the base reference point. Since this principle will be clear from the foregoing description, no further detailed ex planation will be needed for this third embodiment. A fourth embodiment of the operation teaching

method of the invention will be described hereinunder. In teaching the robot the operation using the base

reference points and the comparison reference points, it is essential that the robot be located precisely in relation .to these reference points. Namely, the precision of cor rection of the operation path of the robot is seriously affected by the error in the teaching in relation to the reference points. In order to diminish this error, it is effective to maximize the distance between the refer ence points to be taught. It is also effective to employ a

statistical processing by increasing the number of the reference‘points to be taught. Another effective way for diminishing the error is to limit the precision required at the point to be taught only to a speci?c direction com ponent. Namely, as compared with the indexing of the robot hand end in a three-dimensional space, it is easier to locate the same on a predetermined line (regardless of the position of the direction along the line) and it is still easier to locate the same on a predetermined plane. The fourth embodiment of the invention is based upon this idea, as will be understood from the following descrip tion taken in conjunction with speci?c reference to FIG. 7. In this fourth embodiment, one reference point and one reference plane are used in place of the two reference points employed by the embodiment shown in FIG.'6. More speci?cally, in this embodiment, although the base reference point A1 and the comparison refer ence point B1 are given as in the embodiment shown in FIG. 6, the plane de?ned by the X4 axis and Z4 axis in the coordinate system 04, X4, Y4, ZA and the plane de?ned by the X); axis and Z}; axis in the coordinate system 03, X3, YB, 23 are given, respectively, in place of the base reference point A2 and the comparison refer ence point B2 of the embodiment shown in FIG. 6.

Representing the points to be taught on these planes by A2’ and B2’, respectively, the coordinate values of the point A2’ on the coordinate system 03, X3, Y);, Z]; are expressed as follows, respectively.

O

5

25

40

45

65

12 where kx, hx, a, and B2 are any desired values which can satisfy the conditions of kx=0 and hx=0.

In this case, since the conditions of l,~3=m,3=ni1= n,~g=0 and n,3=l are met, assuming i as being 1 or 2, the transformation matrix TM is given by the following equation (l4).

14 In "111 0 -— m ( )

Tm. : 1.2 "n2 0 -— m 0 0 l — 2K)

0 0 O 1

Since the conditions l,-12+m,-12=I,22+m,22=1 and l,-12+I,22=m,-12+m,z2=l are met, the number of inde pendent unknowns is 4 (four).

Using the equation (13) mentioned before the coordi nate values of the point A1 on the coordinate system 0,4, XA, YA, ZA are expressed as follows.

On the other hand, the coordinate values of the point A2’ on the coordinate system 0,4“, X,4*, YA*, ZA“ can be expressed as follows, using the equation (13).

Since the equation (15.3) affords a condition of X10=Zp1, the value of zloin equation (15.6) can be deter mined regardless of the value of a1.

Equations (15.2) and (15.5) in combination give a condition of mu (ypz —- yp1)=l12 (xp2—xp1). This means that m1; and 112 are not indenpendent from each other, so that 112 and m1; and, hence, l1] and m11 are obtainable. Consequently, x10 and We are determined by the equa tions (15.1) and (15.2), respectively. Thus, all of the unknowns can be determined regardless of the value of kx in the equation (15.4). These facts apply also to the transformation matrix

Twz. It is, therefore, possible to correct the error of the position of the robot, simply by giving one reference point and one reference plane. As will be understood from the foregoing descrip

tion, in giving the reference plane, only one point on such a plane is to be given. Therefore, it is possible to assume a line of intersection between the reference plane and another plane and to assume one point at any desired position on this line of intersection. Namely, the operation teaching method of the invention, in this case, is conducted by using one reference point and one refer ence line.

In the case of welding, the welding line generally coincides with the juncture between two members. Therefore, by using the welding line itself as the refer ence line, it is possible to eliminate the step for teaching the reference line. Namely, the number of the teaching steps can be decreased thanks to the elimination of teaching of the reference line. As an alternative of the method shown in FIG. 5, it is

advisable to conduct the teaching method by giving one reference point, one reference line and one reference

4,613,943 13

plane. It will be clear to those skilled in the art from the foregoing description that the elements of the transfor mation matrix can be determined also in this method. The concepts of the reference line and reference

plane explained hereinbefore are to determine only the necessary ones of the coordinate components of coordi nate values for determining the transformation matrix Twl, Twz, while other coordinate components are se lected to take any desired values, thereby to decrease the number of steps in the teaching operation. Although the invention has been described through

speci?c embodiments applied to a multi-articulated robot R adapted for performing welding, the invention can equally be applied to other types of robots intended for use in other purposes.

In the foregoing description, the teaching of the oper ation path of the robot is effected while the robot is in the ?rst position. This, however, is not exclusive. Namely, when the operation moving path is determined numerically beforehand in accordance with the design data, it is not necessary to teach the operational moving path to the robot while the latter is set in the ?rst posi tion. In this case, the positions of the reference points, line or plane appear optimumly in accordance with the features of the objective structure, e.g. at the corner of the structure.

In the described embodiment, the teaching operation is made by the operator by locating an end of the robot hand at the aimed reference points, lines or planes. However, it will be clear to those skilled in the art that the teaching can be conducted by detecting the infor~ mations for determining the reference points by a visual sensor or a mechanical sensor and automatically com puting the reference points using the detected informa tions. As has been described, according to the invention, it

is possible to effect the correction of the operation path of the robot to eliminate any error which may result from the change in the relative position between the robot and the working objects when the robot is moved to and set at different positions. The invention, how ever, can equally be applied to the correction of the posture of the tool on the robot hand. For instance, this correction can be conducted in a manner explained hereinunder. Namely, in the case of FIG. 2, the posture of the tool on the robot hand when the robot is in the position P is expressed in terms of angular components (Eulerian angle) with respect to the coordinate system 0,4, XA, YA, 2,4 or direction cosines of the tool axis on the axes of the coordinate system, and the tool posture is controlled when the robot is situated at the position Q in such a manner that the angular components with respect to the coordinate system 01;, X3, Y3, Z); or the direction cosines coincide with those obtained when the robot is set at the position P. In this case, the transfor mation of the direction cosines with respect to the coor dinate system 04, X4, Y4, 2,; to those with respect to the coordinate system 03, X3, Y3, ZB can be made by means of the transformation matrices Twl, Twz. Thus, the invention can be applied effectively to such a case that the robot is moved to and set at a plurality of differ ent positions to effect the same operation on similar working objects located at these positions, where a precise control is required not only for the position of an end of the robot hand but also for the posture of the tool held by the robot hand. To sum up, the present invention offers the following

advantages. Namely, it is possible to remarkably

15

25

35

45

55

65

14 shorten the time required for the teaching of the opera tion in the new working position, because the working operation path of the robot for the working object in the new working position can be obtained only simply by showing several comparison reference points at the new position.

Furthermore, according to the invention, it is possi ble to enable the robot to precisely conduct the same operation on a plurality of similar working objects dis posed at different positions, so that the application of the robot can be widened advantageously. What is claimed is: 1. An operation teaching method for an industrial

robot adapted to be successively moved to and set at different positions along an objective structure to con duct a predetermined operation on working objects of the same con?guration on the objective structure to which objects said different positions correspond, re spectively, said method comprising:

teaching, when said industrial robot is set at a ?rst position corresponding to a ?rst working object, positions of base references and a locus of the robot operation assumed on said ?rst working object, and storing such information as coordinate values given in terms of a ?rst robot coordinate system assumed in said robot set insaid ?rst position;

teaching, when said industrial robot is set at a second position corresponding to a second working object, positions of comparison reference assumed on said second working object at positions corresponding to said base references, and storing such informa tion as coordinate values given in terms of a second robot coordinate system assumed in said industrial robot set at said second position;

computing, by using the coordinate values of said base references on said ?rst robot coordinate sys tem, correlation information concerning a base‘ relationship between said ?rst robot coordinate system and a workpiece coordinate system which is determined by said base references;

computing, by using the coordinate values of said comparison references on said second robot coor- _ dinate system, correlation information concerning a comparison relationship between said second robot coordinate system and said workpiece coor dinate system; and

correcting, by using said base relationship and com parison relationship information, the coordinate values representing said locus of the robot opera tion in terms of said ?rst robot coordinate system to determine coordinate values representing said locus of the robot operation in terms of said second robot coordinate system. >

2. An operation teaching method for an industrial robot according to claim 1, wherein said correlation information comprises a transformation matrix having a component corresponding to the amount of transla tional movement of said second robot coordinate sys tem with respect to said ?rst robot coordinate system, and a component corresponding to the direction cosines of said second robot coordinate system with respect to said ?rst robot coordinate system. ‘

3. An operation teaching method for an industrial robot according to claim 2, wherein said base references and said comparison references include at least three different reference informations as reference informa tions for correcting the positional relationship between

4,613,943 15

said second working object and said industrial robot, respectively.

4. An operation teaching method for an industrial robot according to claim 3, wherein said three reference informations comprise a ?rst point assumed on each working object, a second point on a straight line as sumed on each working object and passing said ?rst point but located at a position different from that of said ?rst point, and a third point which is in a plane contain ing said straight line but not existing on said straight line.

5. An operation teaching method for an industrial robot according to claim 2, wherein said base references and said comparison references include two different reference informations as reference informations for correcting the positional relationship between said sec ond working object and said industrial robot.

6. An operation teaching method for an industrial robot according to claim 5, wherein said two reference informations comprise a ?rst point assumed on each working object, and a second point on a straight line assumed on each working object and passing said ?rst point but located at a position different from that of said ?rst point.

7. An operation teaching method for an industrial robot according to claim 5, wherein said two reference informations comprise a ?rst point assumed on each working object, and a second point on a straight line assumed on each working object and passing said ?rst point but located at a position different from that of said ?rst point.

8. An operation teaching method for an industrial robot according to any one of claims 4, 6 and 7 wherein the correction of positional relationship between said second working object and said robot is performed by correcting a coordinate transformation including a translational movement in a three-dimentional space and a rotational movement around a vertical axis or a horizontal axis.

9. An operation teaching method for an industrial robot according to claim 2, wherein said base references and said comparison references include one reference information as a reference information for correcting the positional relationship between said second working object and said industrial robot, respectively.

10. An operation teaching apparatus for an industrial robot adapted to be successively moved to and set at different positions along an objective structure to con duct a predetermined operation on working objects of the same con?guration on said objective structure to which objects said different positions correspond, re spectively, said apparatus comprising: memory means for storing the content of a robot

operation on the working objects taught to said industrial robot;

?rst computing means for obtaining correlation infor mation between coordinate values of base refer ences assumed on a ?rst working object and coor dinate values of comparison references assumed on a second working object, said coordinate values of the base references being stored in said memory means in terms of a ?rst robot coordinate system assumed on said industrial robot set at a ?rst posi tion corresponding to said ?rst working object, said coordinate values of the comparison refer ences being stored in said memory means in terms of a second robot coordinate system assumed on

20

25

45

50

55

60

65

16 said industrial robot in a second position corre sponding to said second working object; and

second computing means for conducting, using said correlation information, correcting computation of coordinate values representing the taught opera tion of the robot and stored in said memory means in terms of said ?rst robot coordinate system to determine coordinate values representing said taught operation in terms of said second robot coordinate system.

11. An operation teaching apparatus for an industrial robot according to claim 10, wherein said ?rst comput ing means includes ?rst means for computing a ?rst transformation matrix which represents correlation information between said ?rst robot coordinate system and a ?rst working object coordinate system assumed on said ?rst working object with respect to said base references; and

second means for computing a second transformation matrix which represents correlation information between said second robot coordinate system and a second working object cordinate system assumed on said second working object with respect to said comparison references, and

third means for computing, using said ?rst and second transformation matrices, a third transformation matrix for correcting the positional error of said second position with respect to said ?rst position.

12. An operation teaching apparatus for an industrial robot according to claim 11, wherein said second com puting means includes means for conducting, using values of said third trans

formation matrix, a correcting computation of said coordinate values representing the taught opera tion of the robot and stored in terms of said ?rst robot coordinate system into the coordinate values representing the taught operation and given in terms of said second robot coordinate system.

13. An operation teaching apparatus for an industrial robot according to claim 12, wherein said base refer ences and said comparison references includes three different reference informations as reference informa tions for correcting the positional relationship between said second working object and said industrial robot, respectively.

14. An operation teaching apparatus for an industrial robot according to claim 13, wherein said three refer ence informations comprise a ?rst point assumed on each working object, a second point on a straight line assumed on each working object and passing said ?rst point but located at a position different from that of said ?rst point, and a third point which is in a plane contain ing said straight line but not existing on said straight line. ,

15. An operation teaching apparatus for an industrial robot according to claim 12, wherein said base refer ences and said comparison references includes two dif ferent reference informations as reference informations for correcting the positional relationship between said second working object and said industrial robot.

16. An operation teaching apparatus for an industrial robot according to claim 15, wherein said two reference informations comprise a ?rst point assumed on each working object, and a second point on a straight line assumed on each working object and passing said ?rst point but located at a position different from that of said ?rst point.

4,613,943 17

17. An operation teaching apparatus for an industrial robot according to claim 15, wherein said two reference informations comprise a reference plane passing through a speci?c axis of each working object coordi nate system parallel to one of the axis of each robot coordinate system, and a reference point assumed on said reference plane.

18. An operation teaching apparatus for an industrial

5

10

15

25

30

35

45

55

65

18 robot according to claim 12, wherein said base refer

ences and said comparison references include one refer

ence information as a reference information for correct

ing the positional relationship between said second working object and said industrial robot.

* * ii * *